Exploring 23andMe's Pharmacogenomic InformationThe FDA website tells us that “genomic biomarkers can play an important role in identifying responders and non-responders, avoiding toxicity and adjusting the dosage of drugs to optimize their efficacy and safety.” Simply put, pharmacogenomics is the study of how genes affect a person’s reaction to drugs, and why different people have different side effects.
The newly appointed director of the National Institutes of Health, Dr. Francis Collins, said that "pharmacogenomics, which from my perspective has been one of the most promising areas of personalized medicine, has also turned out to be extremely complicated, not that we shouldn't have known that." He also said that a blood-thinning drug called Warfarin “has become a poster child for the future of pharmacogenomics."
Looking at the 23andMe Drug Response Section
Warfarin is one of only two drugs (the other being Plavix) that appear under 23andMe’s Drug Response clinical reports section. Another section is called the Research Report. It contains my genetic information on 85 other diseases and traits based on early research that has not qualified yet for the Clinical section. This section lists the only other drug reported on by 23andMe - Statin response. (We’ll look at this section in later posts).
Click on image to enlarge
The report tells me that I have an increased risk for Warfarin sensitivity, which means that if I were ever prescribed Warfarin, my dose should conceivably be slightly less than what is considered typical. The report goes on to tell me that the CYP2C9 and VKORC1 genes are implicated in Warfarin metabolism (which I already know from my work at the Clinical Genetics Institute), and that variants in these genes are consistently associated with Warfarin bleeding complications. A patient’s age, size, other medications, and even diet, are also factored in along with the genotype to compute the most effective dose.
Click on image to enlarge
What I Learned about Warfarin
As I read further into the 23andMe report, I learn that two million people take Warfarin. I wasn’t aware that the number is that high. Blood thinners (or anticoagulants) like Warfarin are given to people at high risk for the formation of blood clots due to conditions such as deep vein thrombosis, irregular heartbeat, and heart valve disease or replacement. The drug is also given to prevent recurrence of pulmonary embolism, heart attack and stroke. Patients taking this drug have regular blood tests to evaluate their blood's clotting ability. The doctor uses the results of these tests to adjust the dose up or down accordingly. One difference I noticed between the Drug Response section of my report and the other sections is that there is no MD’s Perspective tab like here like there are for the diseases and carrier status sections.
Click on image to enlarge
Pharmacogenomic information is contained in about ten percent of labels for drugs approved by the FDA. A significant increase of labels containing such information has been observed over the last decade. To see a list of drugs currently linked to genetics, see the Table of Valid Genomic Biomarkers in the Context of Approved Drug Labels.
A Few Questions
Every genotype variation is either typical sensitivity, or some level of increased sensitivity, which means a decreased dose. Why is it that no genotype requires an INCREASED dose. If it can take weeks of dosing adjustment, what can be the problems or risks a patient might experience during that time? Why would a patient’s age affect a dose?
Don’t Worry Doc – I’m Star One, Star One
The report shows my specific genotype (CYP2C9 *1/*1) that is included in any Warfarin dosing algorithm, of which several are being studied. Under the Resources tab, 23andMe has a link to warfarindosing.org. One 23andMe customer who takes Warfarin entered his genotype and other information into this calculator and found it to be very accurate in predicting his optimal dose.
Click on image to enlarge
The past few years have seen a multitude of pharmacogenomic studies with Warfarin dosing. Intermountain Heathcare is participating in a large multi-center, randomized clinical trial with the NIH that will enroll 1,200 participants. Besides the clinical aspects, one other question being asked in some studies is cost effectiveness. Some results show that Warfarin-related genotyping is unlikely to be cost effective for typical patients with atrial fibrillation, but may be cost effective in patients at high risk for hemorrhage.
Will we have more of this pharmacogenomic info based on or genotype to be informed consumers of medications? Can this work for diet supplements? Will I see any effect from that muscle building shake? And conversely, what fat burning pill will work best for me?
Where’s the pharmacogenomic test for brain function and memory enhancement pills? Or Rogaine effectiveness? Or anti-aging therapies?
Perspective from a medical geneticist
So why did Grant specifically mention Rogaine responsiveness in his last paragraph? It wouldn’t have anything to do with our pictures would it? Look, just because you’re paranoid doesn’t mean they’re not out to get you.
Seriously, before I get to the meat of the discussion let me address one question that Grant raised, “Why is it that no genotype requires an INCREASED dose.” In fact there are patients who are resistant to Warfarin and require very high daily doses. A specific alteration in VKORC1, the same gene that Grant carries a change, has been demonstrated to confer extreme resistance to the effects of Warfarin such that patients require 4-5 times the normal dose to achieve a clinical effect. This is a rare enough occurrence that routine testing for this change has not been proposed, although it is suggested that testing be considered if a patient has severe resistance. There are also non-genetic causes of Warfarin resistance (inability to absorb the drug, medications that interfere with Warfarin action).
Putting the WARF in Warfarin
Warfarin is one of my favorite drugs because it contains a shout out to my alma mater. The ‘WARF’ in Warfarin stands for Wisconsin Alumni Research Foundation, the organization responsible for commercialization of discoveries at the University of Wisconsin-Madison. GO BADGERS! (we now return you to our regularly scheduled blog). I’ve also had a lot of experience with evaluation of the genetic tests for Warfarin dosing, including co-chairing the workgroup of the American College of Medical Genetics (ACMG) that generated an evidence review and guideline for pharmacogenomic testing for Warfarin.
Tricky Drug to Manage
As Grant notes, Warfarin is a very important drug to prevent life-threatening clots in patients with a variety of predispositions. While effective, it is a very tricky drug to manage—too little drug and the patient may clot; too much and they might experience bleeding that can be life-threatening. A number of factors have been identified that affect Warfarin including age, weight, kidney and liver function, other medications and diet. Accounting for all of these factors explains only about 40% of the dose variation from patient to patient. More recently the changes in the genes CYP2C9 and VKORC1 have been shown to explain as much as another 30-40% of the dosing variation. Thus is would seem logical that testing for these genomic variants would be very important to managing this dangerous medication. Certainly the FDA seems to think so.
In the words of college football analyst Lee Corso, “Not so fast my friend”. Let’s examine the question more closely. Glenn Palomaki and Jim Haddow developed an evaluation paradigm for genetic tests called ACCE, where ‘A’ stands for analytic validity, ‘C’ stands for clinical validity, ‘C’ stands for clinical utility and ‘E’ stands for Ethical, Legal and Social issues. For this discussion, we’ll not address ‘E’. Analytic validity addresses the question, does the test measure what it is supposed to measure and is the measurement accurate? There are a number of ways to test for the polymorphisms in the two genes of interest, and the evidence shows that they work extremely well confirming analytic validity. So far so good. The next issue is Clinical Validity which asks, is the test result associated with the ‘disorder’ of interest. In this case, the disorder of interest is Warfarin dose. There is abundant evidence that strongly supports the association of genetic variants in these two genes with ability to predict the stable Warfarin dose.
Here’s where it gets a bit murky. Based on the robust demonstration of Clinical Validity, many began to actively promote testing in the clinical setting. The argument could be summarized as follows: Given that Warfarin is a dangerous drug if dosed improperly and given that genetic testing has been shown to improve dose prediction, it logically follows that genetic testing to inform Warfarin dose will improve patient safety. On the face of it this seems extremely reasonable however there are numerous examples in medicine that refute the idea that reasonable ideas always improve patient care. What are we missing?
This brings us to Clinical Utility which simply stated asks, does the result of the test have an impact on patient outcomes of interest? An outcome is a neutral term that encompasses both positive/desirable and negative/undesirable outcomes. In the case of Warfarin, the positive outcome of interest is achieving an appropriate level of anticoagulation for a given patient. The negative outcomes can be associated with not providing enough anticoagulation (undertreatment) leading to clotting, or providing too much anticoagulation (overtreatment) leading to bleeding. From the patient perspective neither is desirable, that is it’s just as bad to die from a clot as from a bleed.
Because these severe adverse outcomes are relatively infrequent, it is difficult to study these outcomes directly. Researchers fall back on secondary measures (e.g. time to stable dose, time in therapeutic range) that have some evidence that links them to the clinical outcomes of interest. In the case of Warfarin, there are almost no data that address the issue of clinical utility. The evidence review used by the ACMG workgroup to generate its policy statement identified numerous gaps in evidence such that the workgroup’s conclusion (subsequently endorsed by the ACMG leadership) was that there was insufficient evidence to recommend for or against testing. (Click here for the evidence report and policy statement).
The Problem of Bleeding
Another problem is that virtually all the papers addressing this issue focused on the problem of bleeding. As noted above, this is one of two potential negative outcomes and they are two sides of the same coin. To absurdly illustrate this, let’s assume that we want to eliminate all bleeding events caused by Warfarin? The obvious decision would be to not use Warfarin at all. Sure patients would experience clots, but at least they wouldn’t die of bleeds!! Returning to reasonableness requires the acknowledgement that reducing bleeds could increase the risk of clots. The evidence needs to define the relative balance between these two adverse outcomes in order to decide if a strategy has net benefit or net harm. (This problem is not unique to Warfarin. It is disturbing to me that studies consistently focus only on prevention of adverse events even when there is evidence that this could also reduce effectiveness. Subject for another rant…I mean blog).
Intermountain Healthcare Study
To attempt to answer the utility question a prospective randomized clinical trial was conducted at our institution (Intermountain Healthcare) by Dr. Jeff Anderson and his team (reference below). The results of this small trial did not show a net benefit except in certain rare genotypes and interestingly enough in patients with no polymorphisms (so called wild type). It turns out we consistently underdose wild type individuals slightly increasing their risk of clotting. The magnitude of this effect is small, but given that the wild type genotype is the most common, the collective effect may be much larger than anticipated—potentially offsetting the reduction in bleeding events.
My group in conjunction with researchers from the University of Washington used this study to examine the cost-effectiveness of testing, the results of which will appear in the journal Pharmacoeconomics in the near future (authors Meckley, Williams, Gudgeon, Anderson and Veenstra). It is interesting to note that none of the references that raise questions about the evidence for testing, or the Anderson study are listed in the 23andME reference list. Why is this? Are they deliberately trying to hide this information? (Excuse me while I adjust my tin foil hat—the black helicopters are back).
The Economic Issue
What’s the big deal? If it’s possible that we might be doing good why not test, given that the test cost is fairly cheap (estimated at ~$200 clinically). You may have heard something about the health care crisis we have—it’s been in all the papers (Oh right, we don’t read papers any more. Forget I mentioned it). Let’s do a little math. If we accept that 2 million people are on Warfarin and we assume that they all got the test at $200, then we’ve just cost somebody 400 million dollars! If we get no improvement in medical outcomes (or worse we get poorer outcomes), we’ve injected significant cost into the system with no return in improved health. We can no longer afford to behave like this.
Thankfully, others have recognized this issue and there are several large scale prospective studies that are examining genetic testing for Warfarin dosing (include a follow-up study by Dr. Anderson at our institution). Hopefully this means we will know whether or not we should do testing by the time Grant really needs to go on Warfarin!
McClain MR, Palomaki GE, Piper M, Haddow JE. A rapid-ACCE review of CYP2C9 and VKORC1 alleles testing to inform warfarin dosing in adults at elevated risk for thrombotic events to avoid serious bleeding. Genet Med. 2008 Feb;10(2):89-98.
Flockhart DA, O'Kane D, Williams MS, Watson MS, Flockhart DA, Gage B, Gandolfi R, King R, Lyon E, Nussbaum R, O'Kane D, Schulman K, Veenstra D, Williams MS, Watson MS; ACMG Working Group on Pharmacogenetic Testing of CYP2C9, VKORC1 Alleles for Warfarin Use. Pharmacogenetic testing of CYP2C9 and VKORC1 alleles for warfarin. Genet Med. 2008 Feb;10(2):139-50.
Anderson JL, Horne BD, Stevens SM, Grove AS, Barton S, Nicholas ZP, et al. Randomized trial of genotype-guided versus standard warfarin dosing in patients initiating oral anticoagulation. Circulation. 2007 Nov 27;116(22):2563-70.
Meckley LM, Gudgeon JM, Anderson JL, Williams MS, Veenstra DL. A Policy Model to Evaluate the Benefits, Risks, and Costs of Warfarin Pharmacogenomic Testing. Pharmacoeconomics in press.